- Email: [email protected]
Silica-filled polydimethylsiloxane: crystallization process of the adsorbed layer
Silica-filled polydimethylsiloxane: crystallization process of the adsorbed layer D. B o r d e a u x and J. P. C o h e n - A d d a d * Laboratoire de Spectrom6trie Physique associ6 au CNR& Universit~ Joseph Fourier, Grenoble I, BP 87, 38402 St Martin d'Hbres Cedex. France (Received 3 January 1989; revised 28 June 1989; accepted 11 September 1989) Mechanical mixtures of silica particles and molten polydimethylsiloxane give rise to an adsorbed layer. Crystallization properties of this polymeric layer were investigated as a function of the residual amount of polymer Q~ left bound to silica after removing all free chains. The partial crystallinity Pc was determined from the measured enthalpy of crystallization divided by the enthalpy of crystallization per monomer unit. No crystallization effect was observed from very short chains (M, < 6 x 103g moi-1); accordingly, it was considered that loops formed from the adsorption effect are too short to participate in the crystallization. The partial crystallinity Pc associated with longer adsorbed chains was found to vary as 1 - p~ oc Q~-I. It is shown that eight monomeric units, on average, are locked in space by the formation of one H bond between a silanol group and a skeleton oxygen atom. The sol-gel transition resulting from the connection of particles by polymer chains is clearly perceived from the state of crystallinity of the adsorbed layer: when all particles and polymer chains form a gel, the resulting adsorbed layer has a constant partial crystallinity Pc =0.7, whatever the silica concentration and the amount of adsorbed polymer.
(Keywords: polydimethylsiloxane; silica; adsorption; crystallization)
INTRODUCTION This work deals with properties specific to polydimethylsiloxane (PDMS) chains adsorbed upon silica particles that were embedded in systems obtained via mechanical mixing. The elementary process of adsorption of polymeric chains occurs through the hydrogen bonds that the oxygen atoms of the siloxane skeleton form with silanol groups located on the silica surface. It has already been shown that the residual amount of polymer Q, left bound to silica after removing all free chains depends strongly upon the chain molecular weight /d,, on the one hand, and upon the silica concentration Csi on the other 1. The variable Qr was defined per unit mass of silica; it was measured by chemical analysis. More precisely, variations of Q~ within silica-siloxane mixtures corresponding to three concentrations of silica have been investigated over a wide range of chain molecular weights, from M , = 0 . 1 8 x l 0 4 to 35x 104g mol-1. The residual amount Q, has been found to obey the characteristic law:
Q,=zA,£.) "~
(I)
This dependence has been observed within a good accuracy, with Z~=3.3 x 10 -3, 4.3 x 10 -3 and 5.3 × 10- 3 g - 1/2 mop/2, for silica concentrations equal to 0.29, 0.17 and 0.09 w/w, respectively. This experimental result has recently been interpreted by describing any polymer chain adsorbed on the silica surface as a random walk colliding with a plane. Thus, the average number (re) of contact points to the surface can be shown to be proportional to the square root of the number N of skeletal bonds 2: ( r e ) oc N 1/2 * To whom correspondence should be addressed 0032-3861/90/040743-06 © 1990 Butterworth & Co. (Publishers) Ltd.
(2)
Formula (1) is closely related to the Gaussian statistics of polymer chains in a melt. The purpose of the present work was two-fold: (i) We first attempted to determine the average number 7a of monomeric units frozen by the formation of one H bond between one silanol group and one oxygen atom of a chain skeleton. (ii) Attention was then focused upon formulae (1) and (2) by attempting to show that the fraction xa of monomeric units that do not participate in the crystallization process is proportional to (M.)-1/2. The properties of the adsorbed layer were investigated by inducing a crystallization process affecting loops and tails formed from the adsorption process of polymer chains. The principle of this approach was to determine the polymer crystallinity as a function of the amount of polymer Qr left bound to silica after removing all free chains. It was considered that monomeric units bound to or very close to the silica surface have a low mobility and cannot be ordered by lowering the temperature. They cannot participate in the crystallization process.
CHARACTERISTIC F E A T U R E S O F T H E A D S O R P T I O N PROCESS It may be worth emphasizing that the adsorption process resulting from the mechanical mixing of silica particles with siloxane chains in a melt is characterized by several features. Fumed silica consists of particles that obey a mass scaling law; the corresponding fractal exponent, equal to 1.9, has been determined from neutron scattering experiments 3. The average length of particles ranges from 10 a to 3 x 10 a A, while the average diameter may vary from 3 x 10 2 to about 7 x 10 2 A.
POLYMER, 1990, Vol 31, April 743
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
Multiple-chain adsorption Silica particles are fully immersed in a polymer melt, at any time. Consequently, it must be considered that any polymer chain that is in contact with the silica surface is swollen by many other chains, which compete with one another to be adsorbed on the particle: it is a multiple-chain adsorption. Thus, there is a great contrast between this adsorption process and the adsorption occurring on mineral particles in suspension in a polymer solution. Average area associated with one chain The average area #0 associated with the binding of one chain is derived from the specific area A of silica, by assuming that the whole surface participates in the adsorption effect: #c =Ah4./Q~ (3) or
~o=A(~°)I/2/Z~
(4)
So #o is also proportional to (~r),/2. At a silica concentration Cs~=0.09 (w/w), #o is thus found to vary from about 2 x 102 A 2 to about 45 × 102 A 2 within the range of molecular weights 0.18x 104<5~r.<35x I04 g tool- 1 (A = 150 m 2 g- 1). Formulae (2) and (4) are in agreement with each other on assuming that the area associated with one contact point is constant.
Average number of H bonds per chain The average number of free silanol group per unit area of silica has already been estimated4: ve = 1.8 x 10 -2 A -2. This number can be used derive the average size of trains as a function of the chain molecular weight:
#,=~ov°
(5)
/Lt= 1.8 x IO-2A(JVln)I/2/Za
(6)
or
A typical value of/a t is 80 for )~. = 3.4 x 10-5g tool-1
Average thickness The average thickness e of the adsorbed layer is derived from the residual amount of polymer bound to silica and the specific area of silica: ~=Q,/ppa
(7)
is the polymer density. A typical value of a is 133 A for Q,=2gg -1 and A=150m2g -1. Numerical values of ao, #t and ~ should be handled with care; they are only estimates of quantities that determine the binding proces s of one chain. w h e r e pp
Site-bond percolation Mineral particles can be connected to one another when the average size of unstretched chains and the concentration of silica have appropriate values 5. In other words, a site-bond percolation process may occur either by increasing the chain molecular weight and keeping the silica concentration at a constant value, or by increasing the silica concentration and keeping the chain molecular weight constant. The threshold of the site-bond percolation effect is roughly defined by M. = 35 x 104g mol-x at a silica concentration Csi = 0.09 w/w, by M . = 11 x 104gmol -x at Csi=0.23 w/w and by M . = 8 x 104gmo1-1 at Csi=0.29 w/w. Above the threshold of percolation, an infinite cluster is formed, which behaves like a gel: it can be reversibly swollen and it has a modulus of elasticity 6. Below the threshold of gelation, finite clusters are formed: they give rise to a coarse powder. EXPERIMENTAL
Samples Polydimethylsiloxane (PDMS) samples were commercially available polymers. Number-average molecular weights and corresponding polydispersity indices are reported in Table I. Silica-PDMS systems were provided by Rh6ne-Poulenc Recherches (France). Silica-filled polymers were prepared via mechanical mixing. The
Table 1 Calorimetric data concerning pure polymer samples: T is the rate of temperature increase; T8 is the glass transition temperature; To is the cold crystallization temperature; T~, T~ and Ta~ are three melting temperatures observed, AHfa and AH~ denote the enthalpies of fusion; (z~)l and (z~)2 denote the ratios of crystallinity. The index of polydispersity is 1",,2 PDMS M . x 10- 3
T (K rain- 1)
Ts (K)
T~ (K)
Tim (K)
T~ (K)
AH~ (J g- ' )
(z~)z
5.6 5.6 42 44.5 85 250 250 325 325
20 10 20 10 20 20 10 20 10
146 145.5 147 146 147 148 147 149.5 148
199 183 191 190 189 182, 195 191 178 172
227 222 228 229 227.5 228 224 226
234 236 236.5 235 231.5 232.5 231 234
23.0 35.5 33.3 30.5 29.7 30.0 30.4 28.5 30.5
0.30 0.57 0.40 0.39 0.37 0.38 0.38 0.36 0.39
PDMS ~ r x 10 -a
T (K min - t )
Tam (K)
T~ (K)
Tam (K)
AH~ (Jg-~)
('~P)I
5.6 42 44.5 73 85 110 250 325
1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25
231 225 227 226 227 226 227 227
235 238 238 236.5 237 234 235 236
220 236 -
47.9 33.4 33.5 28.7 22.7 25.8 35.0 33.2
0.61 0.42 0.42 0.36 0.51 0.39 0.44 0.42
744
POLYMER, 1990, Vol 31, April
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad Table 2 Calorimetric data concerning the adsorbed layer, in silica-siloxane mixtures: Csi is initial concentration of silica; Qr is residual amount of polymer bound to silica; Pc is normalized ratio of crystallinity; xa is fraction of tightly bound monomeric units; T~ and T2~ are melting temperatures; AHf is enthalpy of fusion; ~sl is ratio of erystallinity without normalization
Cs,
Q,
(g g - 1)
(g g - 1)
1.8
0.29
0.2
0
1
5.6
0.29 0.17 0.09
0.3 0.4 0.5
0 0.15 0.31
1 0,85 0.69
wide wide
42
0.29 0.09
0.7 0.7
0.56 0.63
0,44 0.37
228 228
44.5
0.17
0.8
0.52
0.49
85
0.29 0.09
1.1 1.0
0.68 0.76
0.32 0.32
PDMS J~, x 10- 3
po
T~ (K)
x,,
T~ (K)
Tam (j g- 1) 0
zesi 0
7.4 15
0 0.09 0.19
233 232
18.9 21
0.23 0.26
228.5
232.5
17.1
0.22
230 229
236 234
16.6 18.5
0.21 0.23
110
0.29
1.2
0.67
0.33
227.5
233.5
17.3
0.22
250
0.29 0.17 0.09
1.8 2.3 2.8
0.66 0.65 0.65
0.34 0.35 0.36
228 228 228
232.5 234 234
23.2 22.7 22.6
0.29 0.29 0.29
325
0.05
3.1
0.66
0.34
230
234.5
22
0.28
'l'able 3 Calorimetric data concerning silica-siloxane mixtures obtained after a partial treatment of silica surface: t is fraction of silanol sites converted into non-functional trimethylsilyl groups; Qr is residual amount of polymer; Pc is normalized ratio of crystallinity; x, is fraction of tightly bound monomeric units; T~ and T2~ are melting temperatures; AHf is enthalpy of fusion; zs* is crystallinity of the adsorbed layer PDMS AT, x 10- 3 44.5
73
325
Qr t
(g g- 1)
Pc
xa
0.17 0.50 0.75
0.7 0.33 0.39
0.32 0 0
0,68 . .
0.17 0.22 0.60
1.0 0.9 0.6
0.65 0.68 0.57
0,35 0.33 0,43
0 0.17 0.22 0.60 0.75
2.2 2.0 1.9 1.6 1.0
0.72 0.71 0.67 0.69 0.50
0.28 0.29 0.33 0.31 0.50
T~ (K)
T~ (K)
AH, (j g- 1)
234
10.6
0.13 0 0
wide 229 231
234 234.5
18.5 19.5 16.3
0.23 0.25 0.21
229 229 228 229 228
234.5 234 233 233 233.5
24.0 23.6 22.4 22.8 16.7
0.30 0.30 0.28 0.29 0.21
229 . .
. .
. .
zsi
fumed silica was bought from Degussa (Aerosil, specific area 150m2g-~). The mixtures used throughout this study were characterized by initial concentrations of silica Csi=0.29, 0.17, 0.09 and 0.05 w/w. The residual amount of polymer Qr still bound to silica after removing all free chains was determined from microanalysis measurements of carbon and silicon6. Values of Qr are reported in Table 2. Investigations were extended to mixtures prepared from silica particles obtained after a chemical treatment of the surface. Silica silanol sites were converted into non-functional trimethylsilyl groups according to an experimental procedure described elsewhere 4. Fractions of treatment are reported in Table 3.
consistent thermal treatment was applied to all samples, according to the following conditions. The samples were rapidly cooled to 100 K at a rate of 320 K min- 1. When the system reached thermal equilibrium, the cell was heated to 180 K at a rate of 20 K min-1 and was held at 180K for 30min. Then, it was heated to 280K at a rate of 1.25 K min- t. All experimental results are reported in Table 1. The ratios of crystallinity were calculated from the enthalpy of crystallization per monomer Ahm, determined from the lowering of the melting temperature induced by addition of solvent7; Ahm= 1.4 kcal g- 1.
Differential scanning calorimetry Samples were submitted to thermal cycles by using the cell of a Perkin-Elmer DSC-2 differential scanning calorimeter. The temperature scale and the standardization were calibrated by using cyclohexane as a reference. The calorimeter was adjusted to its highest sensitivity to operate with small amounts of sample (about 10mg). The samples were encapsulated in aluminium pans. The baseline was optimized over the whole experimental temperature range: 100 to 300 K. A
Variations of the partial crystallization %$1 of the adsorbed layer were measured as a function of the variable Qr. They are reported in Figure la. The partial crystallization of the adsorbed layer was also compared with the partial crystallization zP of the corresponding pure polymer melt, measured before any adsorption process 8'9. The ratio Si p__ %/zc - Pc is called the normalized crystallinity; variations of Pc are reported in Figure lb. The striking feature perceived from Figure I concerns the three well defined ranges determined from the variations of ~csi or Pc.
PARTIAL CRYSTALLIZATION OF THE ADSORBED LAYER
POLYMER, 1990, Vol 31, April
745
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
Partial crystallinity
°I
v~
&
b
0"6f
0.4
~°
-_
0.2
~
==.=1
Assuming that the whole area of silica particles participates in the adsorption process, the total number of bonds formed with silanol groups is A/(r=, where A is the specific area of silica and tr= is the average area associated with a silanol site: ire=55/~2. Furthermore, it is considered that the number of monomeric units locked by the effect of adsorption is Y, per H bond. Then, the total number of locked monomer units is AT~/a~ per gram of silica. The fraction of bonds participating in the crystallization process is: Xa =
A,-,r
0
J
I
1.0
0.5
I
I
I
I
1.5
2.0
2.5
3.0
O R ( w / w l
Figure 1 Partial crystallinity of the adsorbed layer. Measurements were performed upon silica-siloxane mixtures after removing all free chains; Q, is the residual amount.of polymer bound to silica. (a) The partial crystallinity ratio zsi observed from bound PDMS chains. (b) The partial crystalhmty Pc normahzed with respect to the erystallinity of the corresponding pure polymer. Number-average molecular weights (10-'M.) are: 32.5 (A), 25 (A), (.), 8.5 (V), 4.2 (V), 0.56 (I-q) and 0.18 (0). All silica concentrations are reported in Table 2 •
.
¢
,
.
11
Considering the normalized curve in Figure la, no crystallization (Pc = 0) was observed within the accuracy of the experiments for Qr values lower than 0.3 g g - 1 ; then a sharp increase of the partial crystallization pC, from 0 to 0.7, was observed within the range 0.3 < Qr < 1.0gg-1; finally, a plateau value 0.7 was observed for Pc, when Q~ varies from 1.1 to 3.1gg -1. It is worth emphasizing that silica-siloxane mixtures corresponding to values of Q, higher than 1.0 exhibit a gel behaviour. Consequently, the transition from the sol, where finite clusters are formed, to the gel, where silica particles are connected to one another by chains, is perceived from the state of partial crystallinity of the adsorbed layer. A D S O R P T I O N OF SHORT CHAINS
1 -- Pc =
fflmATJtreQ,
(8)
where ~t m is the molecular weight of one monomeric unit. This fraction must be a linear function of Q71 Variations of the experimental quantity 1 - Pc are reported in Figure 2 as a function of the inverse of the adsorbed amount of polymer. The experimental value of the slope of the straight line drawn in Figure 2 is 0.35 g g-1. The numerical value of 7, calculated from formula (8) is then equal to 10. This value is in reasonable agreement with the average number obtained in the foregoing section and corresponding to frozen monomer units associated with one H bond formed with a silanol group. The comparison of the two results only shows that the adsorption of short chains does not correspond to the maximum length of loops that are locked as a result of the binding process.
Law of adsorption A direct experimental approach to the law of adsorption given by formula (1) can be obtained by representing the variations of the inverse of the fraction x, = 1 - P c as a function of (M,) 1/2. The corresponding curve is reported in Figure 3; it is a straight line. The silica concentration was Csi = 0.09 (w/w). In accordance with the experimental law (1) and the interpretation given in ref. 2, the average number of contact points of one chain with the silica surface is expressed as:
=e.(N,) 1/2
(9)
Starting from an initial concentration of silica Csi equal to 0.29 (w/w), two chain lengths were found to give rise to the value pc=0; they correspond to ~,t.= 1.8 x 103 and M , = 5 . 6 x 10agmol -1. They are associated with residual amounts of polymer equal to 0.23 and 0.30 g g - 1, respectively. It is considered that the absence of crystallization shows that all loops that are formed during the adsorption process are too short to participate in the crystallization process. In that case, the fraction x, of mononieric units tightly bound to the surface is equal to unity. The average number of chains adsorbed on the particles is 7.7 x 1019 for Qr=0.23gg-1 and 3.2 x 1019 for Qr = 0.3 g g-1, while the number of monomer units frozen by the adsorption effect is 19 x 1020 in the first case and 24 x 1020 in the second one. The number of H bonds available per gram of silica is 2.7 x 102°. Accordingly, about eight monomer units were frozen by the formation of one H bond between one monomer unit of the chain skeleton and one silanol group of the surface.
o.s
FINITE CLUSTERS
Figure 2 :Fbefraction x== 1-po of monomeric units tightly bound
A strong increase of the normalized crystallinity was observed within the range 0.3 g g-1 ~
746
POLYMER,1990,Vol 31, April
where N, is. the number average of bonds in one chain 1.0
08
[]S
0.4 o.z ~ ' Q ' 0
I
0.4. I
I
0.8 I
I
1.2 1.6 OR Ilwlwl I
I
I
I
2.O I
I
2.4 I
to the silica surface represented as a function of the inverse of the residual amount of polymer. Key to symbols is given in Figure 1. Measurements corresponding to treated silica are also reported (Ak within <> and ©)
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
3
0
l
I
I
I00
200
300
Figure 3 The inverse of the fraction of tightly bound monomeric units reported as a function of the square root of the chain molecular weight (the silica concentration is Csi=0.29 w/w)
and e~ is a numerical factor depending upon the silica concentration. Then, the residual amount Q~ of polymer bound to silica is:
Qr = [A(Nlm)l/2/tre%~Cl/2"](ffi.) 1/2
(10)
partial crystallization of the adsorbed layer. It is worth emphasizing that the site-bond percolation effect is well understood from the ordering properties of the adsorbed layer. Although the amount of polymer Qr confined between particles varies from 1.0 to 3.1 g g- 1, the fraction Xa of monomeric units tightly bound to silica has a constant value. This is also observed by keeping the number-average chain molecular weight constant and varying the silica concentration. It is not the purpose of the present work to give a thorough description of all properties of the adsorbed layer. The observation of a plateau associated with the formation of a gel can be discussed in the following way. (i) It is first noticed that the law of adsorption (1) concerns all polymer samples studied from the partial crystallization of the adsorbed layer. Consequently, neither the specific area A of silica nor the parameter % must be considered as varying with Q, within the range corresponding to the plateau (formulae (10) and (11)). (ii) Then, considering formula (8) or (12), it is assumed that the parameter Ya may depend upon the amount of polymer bound to silica when a gel is formed. Experimentally, it is found that 7~must be proportional to Qr. The residual amount Qr is increased either by lowering the silica concentration or by lengthening polymer chains. It could be considered that in both cases PDMS chains are given more freedom when Qr is increased; consequently, they should be more easily ordered during the crystallization process. This is contrary to the experimental result. Thus, it is assumed that additional constraints arise from the increase of Qr; these constraints may be identified as entanglements, for example. This will be discussed in a subsequent work.
Therefore: X a -~- A ( I V l m ) l / 2 / a e g a ~ l / 2
(11)
where d is the Avogadro number. Correspondingly, the fraction xa of monomeric units locked by the adsorption process is expressed as: x. = %7~(Mm)l/2/(M,) 1/2
(12)
The inverse of the fraction x~ must vary as a linear function of (~.)1/2. The slope of the straight line drawn in Figure 3 leads to a numerical value of the product eaT~ equal to 1.4. Considering that Y~- 10, the value of ea is about 0.14. This estimate of % accounts for stiffness effects concerning real PDMS chains; it represents the deviation from a pure random walk. INFINITE CLUSTERS: SILICA-SILOXANE GELS When the concentration of silica Csi and the chain molecular weight M, reach values higher than those corresponding to the threshold of site-bond percolation, an infinite cluster is formed. Silica aggregates are connected to one another by PDMS chains. The resulting systems can be reversibly swollen by using a good solvent. They also exhibit an elasticity property that can be characterized by a modulus. Swelling properties have already been described elsewhere 6. The formation of a silica-siloxane gel can be experimentally associated with the range of residual amount of polymer Q, going from 1.0 to 3.1 g g - 1. It is clearly seen from Figure I that this range corresponds to a plateau value of the normalized
TREATED SILICA SURFACE Finally, the threshold of percolation was shifted towards higher silica concentrations at constant chain length or towards higher chain lengths at constant silica concentration by treating the silica surface. Hexamethyldisilazane was used to deactivate a partial number of silanol groups.
Finite clusters One of the most striking results concerns the mixture made from polymer chains of number-average molecular weight M n = 44.5 x 10a g mol- 1. The initial concentration of the mixture was Csi=0.17 (w/w). In the absence of any treatment of the silica surface, the residual amount of polymer Qr left bound to silica was 0.8 gg-1. Then, the values of Qr were 0.7, 0.4 and 0.4 g g- 1 for converted free silanol fractions equal to 0.17, 0.72 and 0.75, respectively. The corresponding partial crystallinity ratios were pc=0.32, 0 and 0, respectively (Table 3, Figure 4). Therefore, it is considered that loops formed from the adsorption process can collapse onto the silica surface when the density of free silanol groups is smaller than 0.3. Accordingly, they cannot participate in the crystallization process. Mixtures made from polymer chains characterized by a number-average molecular weight M n = 7.3 x 104 were also observed. Ratios of converted silanols were 0.17, 0.22 and 0.60; corresponding amounts of residual polymer were 1.0, 0.9 and 0.6 g g- 1, respectively. The partial crystallinity was found to be equal to 0.65, 0.68 and 0.57 (Table 3, Figure 4). Corresponding fractions of frozen bonds are reported in Figure 2.
POLYMER, 1990, Vol 31, April
747
Crystallization of s/l/ca-filled PDMS: D. Bordeaux and J. P. Cohen-Addad 1.0 0.8
~
nn
nl
N
b
0.6
m 0.4 ifiu
o
0.2 I
0
I
1.5 2.0 OR(W/W)
I
I
2.5
5,0
Figure 4 The partial crystallinity observed from mixtures of siloxane and silica with a chemical treatment of the surface. Experimental points are compared with the average curves obtained without any chemical treatment (Figure I). (a) The ratio of crystallinity ~si observed on PDMS chains. (b) The crystallinity Pc normalized with respect to that of corresponding pure polymer chains. Number-average chain molecular weights (10-4_~r,) are: 32.5 (4, within r-q), 7.3 (4, within <>) and 4.45 (4, within C)). Corresponding points without any treatment are also reported in Figure 1. Fractions of treatment are reported in Table 3
Infinite cluster Treated silica was also used to observe the formation of infinite clusters. The concentration of silica was kept constant, Csi=0.17(w/w), and the chain molecular weight was M, = 3.25 x 105. As long as an infinite cluster was formed, the partial crystallinity was characterized by pc_0.7. Then, for Q , < I , the partial crystallinity was found to decrease. Figures 1 and 4 clearly show that the crystallinity ratio is a constant in all silica-siloxane mixtures that form a gel although the silica concentration and the chain length may vary from one system to another. The fraction of trains (or of loops and tails) has a constant value whatever the amount of polymer bound per gram of silica. This property will also be discussed in a subsequent work. CONCLUSIONS The mechanical mixing of siloxane chains and silica particles induces an adsorption process for which char-
748
POLYMER, 1990, Vol 31, April
acteristic properties arise mainly from the full immersion effect of particles in the polymer melt: the amount of adsorbed polymer is proportional to the square root of the number-average molecular weight of chains. The purpose of the present work was to give a preliminary investigation of the structure of the adsorbed layers. The degree of freedom of monomeric units is determined by observing the partial crystallinity of the polymeric phase. Differential scanning calorimetry measurements performed after removing all free chains showed that about 10 monomeric units are frozen by the formation of one hydrogen bond between one silanol group and one oxygen atom of a chain skeleton. The site--bond percolation induced by polymer chains connecting silica particles was perceived from the state of partial crystallinity of the adsorbed layer. This preliminary study will be extended by an n.m.r. approach to the kinetics of crystallization of the polymeric layer 7. Measurements of differential scanning calorimetry performed on the adsorbed layer of silica-siloxane mixtures show clear evidence for the existence of a site-bond percolation transition of silica particles. The percolation process occurs through polymer molecules. The present work also shows that eight monomeric units on average are fixed in space by the formation of one hydrogen bond between a silanol group and a skeleton oxygen atom. The analysis of experimental results suggests that the area of silica surface available to polymer chains might vary the silica concentration. REFERENCES 1 Cohen-Addad, J. P., Roby, C. and Sauviat, M. Polymer 1985, 26, 1231 2 Cohen-Addad, J. P. Polymer 1989, 30, 1820 3 Schaefer, D. W. MRS Bull. 1988, XIII, 222 4 Viallat, A., Cohen-Addad, J. P. and Pouchleon, A, Polymer 1986, 27, 843 5 Stauffer, D. 'Introduction to Percolation Theory', Taylor and Francis, London, 1985 6 Cohen-Addad, J. P., Viallat, A. and Huchot, P. Macromolecules 1987, 20, 2146 7 Feio, G., Buntinx, G. and Cohen-Addad, J. P. J. Polym. Sci. 1989, 27, 1 8 Orrah, D. J., Semlyen, J. A., Dodgson, K. and Ross-Murphy, S. B. Polymer 1987, 28, 985 9 Clarson, J., Mark, J. E. and Dodgson, K. Polym. Commun. 1988, 29, 208
(Keywords: polydimethylsiloxane; silica; adsorption; crystallization)
INTRODUCTION This work deals with properties specific to polydimethylsiloxane (PDMS) chains adsorbed upon silica particles that were embedded in systems obtained via mechanical mixing. The elementary process of adsorption of polymeric chains occurs through the hydrogen bonds that the oxygen atoms of the siloxane skeleton form with silanol groups located on the silica surface. It has already been shown that the residual amount of polymer Q, left bound to silica after removing all free chains depends strongly upon the chain molecular weight /d,, on the one hand, and upon the silica concentration Csi on the other 1. The variable Qr was defined per unit mass of silica; it was measured by chemical analysis. More precisely, variations of Q~ within silica-siloxane mixtures corresponding to three concentrations of silica have been investigated over a wide range of chain molecular weights, from M , = 0 . 1 8 x l 0 4 to 35x 104g mol-1. The residual amount Q, has been found to obey the characteristic law:
Q,=zA,£.) "~
(I)
This dependence has been observed within a good accuracy, with Z~=3.3 x 10 -3, 4.3 x 10 -3 and 5.3 × 10- 3 g - 1/2 mop/2, for silica concentrations equal to 0.29, 0.17 and 0.09 w/w, respectively. This experimental result has recently been interpreted by describing any polymer chain adsorbed on the silica surface as a random walk colliding with a plane. Thus, the average number (re) of contact points to the surface can be shown to be proportional to the square root of the number N of skeletal bonds 2: ( r e ) oc N 1/2 * To whom correspondence should be addressed 0032-3861/90/040743-06 © 1990 Butterworth & Co. (Publishers) Ltd.
(2)
Formula (1) is closely related to the Gaussian statistics of polymer chains in a melt. The purpose of the present work was two-fold: (i) We first attempted to determine the average number 7a of monomeric units frozen by the formation of one H bond between one silanol group and one oxygen atom of a chain skeleton. (ii) Attention was then focused upon formulae (1) and (2) by attempting to show that the fraction xa of monomeric units that do not participate in the crystallization process is proportional to (M.)-1/2. The properties of the adsorbed layer were investigated by inducing a crystallization process affecting loops and tails formed from the adsorption process of polymer chains. The principle of this approach was to determine the polymer crystallinity as a function of the amount of polymer Qr left bound to silica after removing all free chains. It was considered that monomeric units bound to or very close to the silica surface have a low mobility and cannot be ordered by lowering the temperature. They cannot participate in the crystallization process.
CHARACTERISTIC F E A T U R E S O F T H E A D S O R P T I O N PROCESS It may be worth emphasizing that the adsorption process resulting from the mechanical mixing of silica particles with siloxane chains in a melt is characterized by several features. Fumed silica consists of particles that obey a mass scaling law; the corresponding fractal exponent, equal to 1.9, has been determined from neutron scattering experiments 3. The average length of particles ranges from 10 a to 3 x 10 a A, while the average diameter may vary from 3 x 10 2 to about 7 x 10 2 A.
POLYMER, 1990, Vol 31, April 743
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
Multiple-chain adsorption Silica particles are fully immersed in a polymer melt, at any time. Consequently, it must be considered that any polymer chain that is in contact with the silica surface is swollen by many other chains, which compete with one another to be adsorbed on the particle: it is a multiple-chain adsorption. Thus, there is a great contrast between this adsorption process and the adsorption occurring on mineral particles in suspension in a polymer solution. Average area associated with one chain The average area #0 associated with the binding of one chain is derived from the specific area A of silica, by assuming that the whole surface participates in the adsorption effect: #c =Ah4./Q~ (3) or
~o=A(~°)I/2/Z~
(4)
So #o is also proportional to (~r),/2. At a silica concentration Cs~=0.09 (w/w), #o is thus found to vary from about 2 x 102 A 2 to about 45 × 102 A 2 within the range of molecular weights 0.18x 104<5~r.<35x I04 g tool- 1 (A = 150 m 2 g- 1). Formulae (2) and (4) are in agreement with each other on assuming that the area associated with one contact point is constant.
Average number of H bonds per chain The average number of free silanol group per unit area of silica has already been estimated4: ve = 1.8 x 10 -2 A -2. This number can be used derive the average size of trains as a function of the chain molecular weight:
#,=~ov°
(5)
/Lt= 1.8 x IO-2A(JVln)I/2/Za
(6)
or
A typical value of/a t is 80 for )~. = 3.4 x 10-5g tool-1
Average thickness The average thickness e of the adsorbed layer is derived from the residual amount of polymer bound to silica and the specific area of silica: ~=Q,/ppa
(7)
is the polymer density. A typical value of a is 133 A for Q,=2gg -1 and A=150m2g -1. Numerical values of ao, #t and ~ should be handled with care; they are only estimates of quantities that determine the binding proces s of one chain. w h e r e pp
Site-bond percolation Mineral particles can be connected to one another when the average size of unstretched chains and the concentration of silica have appropriate values 5. In other words, a site-bond percolation process may occur either by increasing the chain molecular weight and keeping the silica concentration at a constant value, or by increasing the silica concentration and keeping the chain molecular weight constant. The threshold of the site-bond percolation effect is roughly defined by M. = 35 x 104g mol-x at a silica concentration Csi = 0.09 w/w, by M . = 11 x 104gmol -x at Csi=0.23 w/w and by M . = 8 x 104gmo1-1 at Csi=0.29 w/w. Above the threshold of percolation, an infinite cluster is formed, which behaves like a gel: it can be reversibly swollen and it has a modulus of elasticity 6. Below the threshold of gelation, finite clusters are formed: they give rise to a coarse powder. EXPERIMENTAL
Samples Polydimethylsiloxane (PDMS) samples were commercially available polymers. Number-average molecular weights and corresponding polydispersity indices are reported in Table I. Silica-PDMS systems were provided by Rh6ne-Poulenc Recherches (France). Silica-filled polymers were prepared via mechanical mixing. The
Table 1 Calorimetric data concerning pure polymer samples: T is the rate of temperature increase; T8 is the glass transition temperature; To is the cold crystallization temperature; T~, T~ and Ta~ are three melting temperatures observed, AHfa and AH~ denote the enthalpies of fusion; (z~)l and (z~)2 denote the ratios of crystallinity. The index of polydispersity is 1",,2 PDMS M . x 10- 3
T (K rain- 1)
Ts (K)
T~ (K)
Tim (K)
T~ (K)
AH~ (J g- ' )
(z~)z
5.6 5.6 42 44.5 85 250 250 325 325
20 10 20 10 20 20 10 20 10
146 145.5 147 146 147 148 147 149.5 148
199 183 191 190 189 182, 195 191 178 172
227 222 228 229 227.5 228 224 226
234 236 236.5 235 231.5 232.5 231 234
23.0 35.5 33.3 30.5 29.7 30.0 30.4 28.5 30.5
0.30 0.57 0.40 0.39 0.37 0.38 0.38 0.36 0.39
PDMS ~ r x 10 -a
T (K min - t )
Tam (K)
T~ (K)
Tam (K)
AH~ (Jg-~)
('~P)I
5.6 42 44.5 73 85 110 250 325
1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25
231 225 227 226 227 226 227 227
235 238 238 236.5 237 234 235 236
220 236 -
47.9 33.4 33.5 28.7 22.7 25.8 35.0 33.2
0.61 0.42 0.42 0.36 0.51 0.39 0.44 0.42
744
POLYMER, 1990, Vol 31, April
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad Table 2 Calorimetric data concerning the adsorbed layer, in silica-siloxane mixtures: Csi is initial concentration of silica; Qr is residual amount of polymer bound to silica; Pc is normalized ratio of crystallinity; xa is fraction of tightly bound monomeric units; T~ and T2~ are melting temperatures; AHf is enthalpy of fusion; ~sl is ratio of erystallinity without normalization
Cs,
Q,
(g g - 1)
(g g - 1)
1.8
0.29
0.2
0
1
5.6
0.29 0.17 0.09
0.3 0.4 0.5
0 0.15 0.31
1 0,85 0.69
wide wide
42
0.29 0.09
0.7 0.7
0.56 0.63
0,44 0.37
228 228
44.5
0.17
0.8
0.52
0.49
85
0.29 0.09
1.1 1.0
0.68 0.76
0.32 0.32
PDMS J~, x 10- 3
po
T~ (K)
x,,
T~ (K)
Tam (j g- 1) 0
zesi 0
7.4 15
0 0.09 0.19
233 232
18.9 21
0.23 0.26
228.5
232.5
17.1
0.22
230 229
236 234
16.6 18.5
0.21 0.23
110
0.29
1.2
0.67
0.33
227.5
233.5
17.3
0.22
250
0.29 0.17 0.09
1.8 2.3 2.8
0.66 0.65 0.65
0.34 0.35 0.36
228 228 228
232.5 234 234
23.2 22.7 22.6
0.29 0.29 0.29
325
0.05
3.1
0.66
0.34
230
234.5
22
0.28
'l'able 3 Calorimetric data concerning silica-siloxane mixtures obtained after a partial treatment of silica surface: t is fraction of silanol sites converted into non-functional trimethylsilyl groups; Qr is residual amount of polymer; Pc is normalized ratio of crystallinity; x, is fraction of tightly bound monomeric units; T~ and T2~ are melting temperatures; AHf is enthalpy of fusion; zs* is crystallinity of the adsorbed layer PDMS AT, x 10- 3 44.5
73
325
Qr t
(g g- 1)
Pc
xa
0.17 0.50 0.75
0.7 0.33 0.39
0.32 0 0
0,68 . .
0.17 0.22 0.60
1.0 0.9 0.6
0.65 0.68 0.57
0,35 0.33 0,43
0 0.17 0.22 0.60 0.75
2.2 2.0 1.9 1.6 1.0
0.72 0.71 0.67 0.69 0.50
0.28 0.29 0.33 0.31 0.50
T~ (K)
T~ (K)
AH, (j g- 1)
234
10.6
0.13 0 0
wide 229 231
234 234.5
18.5 19.5 16.3
0.23 0.25 0.21
229 229 228 229 228
234.5 234 233 233 233.5
24.0 23.6 22.4 22.8 16.7
0.30 0.30 0.28 0.29 0.21
229 . .
. .
. .
zsi
fumed silica was bought from Degussa (Aerosil, specific area 150m2g-~). The mixtures used throughout this study were characterized by initial concentrations of silica Csi=0.29, 0.17, 0.09 and 0.05 w/w. The residual amount of polymer Qr still bound to silica after removing all free chains was determined from microanalysis measurements of carbon and silicon6. Values of Qr are reported in Table 2. Investigations were extended to mixtures prepared from silica particles obtained after a chemical treatment of the surface. Silica silanol sites were converted into non-functional trimethylsilyl groups according to an experimental procedure described elsewhere 4. Fractions of treatment are reported in Table 3.
consistent thermal treatment was applied to all samples, according to the following conditions. The samples were rapidly cooled to 100 K at a rate of 320 K min- 1. When the system reached thermal equilibrium, the cell was heated to 180 K at a rate of 20 K min-1 and was held at 180K for 30min. Then, it was heated to 280K at a rate of 1.25 K min- t. All experimental results are reported in Table 1. The ratios of crystallinity were calculated from the enthalpy of crystallization per monomer Ahm, determined from the lowering of the melting temperature induced by addition of solvent7; Ahm= 1.4 kcal g- 1.
Differential scanning calorimetry Samples were submitted to thermal cycles by using the cell of a Perkin-Elmer DSC-2 differential scanning calorimeter. The temperature scale and the standardization were calibrated by using cyclohexane as a reference. The calorimeter was adjusted to its highest sensitivity to operate with small amounts of sample (about 10mg). The samples were encapsulated in aluminium pans. The baseline was optimized over the whole experimental temperature range: 100 to 300 K. A
Variations of the partial crystallization %$1 of the adsorbed layer were measured as a function of the variable Qr. They are reported in Figure la. The partial crystallization of the adsorbed layer was also compared with the partial crystallization zP of the corresponding pure polymer melt, measured before any adsorption process 8'9. The ratio Si p__ %/zc - Pc is called the normalized crystallinity; variations of Pc are reported in Figure lb. The striking feature perceived from Figure I concerns the three well defined ranges determined from the variations of ~csi or Pc.
PARTIAL CRYSTALLIZATION OF THE ADSORBED LAYER
POLYMER, 1990, Vol 31, April
745
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
Partial crystallinity
°I
v~
&
b
0"6f
0.4
~°
-_
0.2
~
==.=1
Assuming that the whole area of silica particles participates in the adsorption process, the total number of bonds formed with silanol groups is A/(r=, where A is the specific area of silica and tr= is the average area associated with a silanol site: ire=55/~2. Furthermore, it is considered that the number of monomeric units locked by the effect of adsorption is Y, per H bond. Then, the total number of locked monomer units is AT~/a~ per gram of silica. The fraction of bonds participating in the crystallization process is: Xa =
A,-,r
0
J
I
1.0
0.5
I
I
I
I
1.5
2.0
2.5
3.0
O R ( w / w l
Figure 1 Partial crystallinity of the adsorbed layer. Measurements were performed upon silica-siloxane mixtures after removing all free chains; Q, is the residual amount.of polymer bound to silica. (a) The partial crystallinity ratio zsi observed from bound PDMS chains. (b) The partial crystalhmty Pc normahzed with respect to the erystallinity of the corresponding pure polymer. Number-average molecular weights (10-'M.) are: 32.5 (A), 25 (A), (.), 8.5 (V), 4.2 (V), 0.56 (I-q) and 0.18 (0). All silica concentrations are reported in Table 2 •
.
¢
,
.
11
Considering the normalized curve in Figure la, no crystallization (Pc = 0) was observed within the accuracy of the experiments for Qr values lower than 0.3 g g - 1 ; then a sharp increase of the partial crystallization pC, from 0 to 0.7, was observed within the range 0.3 < Qr < 1.0gg-1; finally, a plateau value 0.7 was observed for Pc, when Q~ varies from 1.1 to 3.1gg -1. It is worth emphasizing that silica-siloxane mixtures corresponding to values of Q, higher than 1.0 exhibit a gel behaviour. Consequently, the transition from the sol, where finite clusters are formed, to the gel, where silica particles are connected to one another by chains, is perceived from the state of partial crystallinity of the adsorbed layer. A D S O R P T I O N OF SHORT CHAINS
1 -- Pc =
fflmATJtreQ,
(8)
where ~t m is the molecular weight of one monomeric unit. This fraction must be a linear function of Q71 Variations of the experimental quantity 1 - Pc are reported in Figure 2 as a function of the inverse of the adsorbed amount of polymer. The experimental value of the slope of the straight line drawn in Figure 2 is 0.35 g g-1. The numerical value of 7, calculated from formula (8) is then equal to 10. This value is in reasonable agreement with the average number obtained in the foregoing section and corresponding to frozen monomer units associated with one H bond formed with a silanol group. The comparison of the two results only shows that the adsorption of short chains does not correspond to the maximum length of loops that are locked as a result of the binding process.
Law of adsorption A direct experimental approach to the law of adsorption given by formula (1) can be obtained by representing the variations of the inverse of the fraction x, = 1 - P c as a function of (M,) 1/2. The corresponding curve is reported in Figure 3; it is a straight line. The silica concentration was Csi = 0.09 (w/w). In accordance with the experimental law (1) and the interpretation given in ref. 2, the average number of contact points of one chain with the silica surface is expressed as:
(9)
Starting from an initial concentration of silica Csi equal to 0.29 (w/w), two chain lengths were found to give rise to the value pc=0; they correspond to ~,t.= 1.8 x 103 and M , = 5 . 6 x 10agmol -1. They are associated with residual amounts of polymer equal to 0.23 and 0.30 g g - 1, respectively. It is considered that the absence of crystallization shows that all loops that are formed during the adsorption process are too short to participate in the crystallization process. In that case, the fraction x, of mononieric units tightly bound to the surface is equal to unity. The average number of chains adsorbed on the particles is 7.7 x 1019 for Qr=0.23gg-1 and 3.2 x 1019 for Qr = 0.3 g g-1, while the number of monomer units frozen by the adsorption effect is 19 x 1020 in the first case and 24 x 1020 in the second one. The number of H bonds available per gram of silica is 2.7 x 102°. Accordingly, about eight monomer units were frozen by the formation of one H bond between one monomer unit of the chain skeleton and one silanol group of the surface.
o.s
FINITE CLUSTERS
Figure 2 :Fbefraction x== 1-po of monomeric units tightly bound
A strong increase of the normalized crystallinity was observed within the range 0.3 g g-1 ~
746
POLYMER,1990,Vol 31, April
where N, is. the number average of bonds in one chain 1.0
08
[]S
0.4 o.z ~ ' Q ' 0
I
0.4. I
I
0.8 I
I
1.2 1.6 OR Ilwlwl I
I
I
I
2.O I
I
2.4 I
to the silica surface represented as a function of the inverse of the residual amount of polymer. Key to symbols is given in Figure 1. Measurements corresponding to treated silica are also reported (Ak within <> and ©)
Crystallization of silica-filled PDMS: D. Bordeaux and J. P. Cohen-Addad
3
0
l
I
I
I00
200
300
Figure 3 The inverse of the fraction of tightly bound monomeric units reported as a function of the square root of the chain molecular weight (the silica concentration is Csi=0.29 w/w)
and e~ is a numerical factor depending upon the silica concentration. Then, the residual amount Q~ of polymer bound to silica is:
Qr = [A(Nlm)l/2/tre%~Cl/2"](ffi.) 1/2
(10)
partial crystallization of the adsorbed layer. It is worth emphasizing that the site-bond percolation effect is well understood from the ordering properties of the adsorbed layer. Although the amount of polymer Qr confined between particles varies from 1.0 to 3.1 g g- 1, the fraction Xa of monomeric units tightly bound to silica has a constant value. This is also observed by keeping the number-average chain molecular weight constant and varying the silica concentration. It is not the purpose of the present work to give a thorough description of all properties of the adsorbed layer. The observation of a plateau associated with the formation of a gel can be discussed in the following way. (i) It is first noticed that the law of adsorption (1) concerns all polymer samples studied from the partial crystallization of the adsorbed layer. Consequently, neither the specific area A of silica nor the parameter % must be considered as varying with Q, within the range corresponding to the plateau (formulae (10) and (11)). (ii) Then, considering formula (8) or (12), it is assumed that the parameter Ya may depend upon the amount of polymer bound to silica when a gel is formed. Experimentally, it is found that 7~must be proportional to Qr. The residual amount Qr is increased either by lowering the silica concentration or by lengthening polymer chains. It could be considered that in both cases PDMS chains are given more freedom when Qr is increased; consequently, they should be more easily ordered during the crystallization process. This is contrary to the experimental result. Thus, it is assumed that additional constraints arise from the increase of Qr; these constraints may be identified as entanglements, for example. This will be discussed in a subsequent work.
Therefore: X a -~- A ( I V l m ) l / 2 / a e g a ~ l / 2
(11)
where d is the Avogadro number. Correspondingly, the fraction xa of monomeric units locked by the adsorption process is expressed as: x. = %7~(Mm)l/2/(M,) 1/2
(12)
The inverse of the fraction x~ must vary as a linear function of (~.)1/2. The slope of the straight line drawn in Figure 3 leads to a numerical value of the product eaT~ equal to 1.4. Considering that Y~- 10, the value of ea is about 0.14. This estimate of % accounts for stiffness effects concerning real PDMS chains; it represents the deviation from a pure random walk. INFINITE CLUSTERS: SILICA-SILOXANE GELS When the concentration of silica Csi and the chain molecular weight M, reach values higher than those corresponding to the threshold of site-bond percolation, an infinite cluster is formed. Silica aggregates are connected to one another by PDMS chains. The resulting systems can be reversibly swollen by using a good solvent. They also exhibit an elasticity property that can be characterized by a modulus. Swelling properties have already been described elsewhere 6. The formation of a silica-siloxane gel can be experimentally associated with the range of residual amount of polymer Q, going from 1.0 to 3.1 g g - 1. It is clearly seen from Figure I that this range corresponds to a plateau value of the normalized
TREATED SILICA SURFACE Finally, the threshold of percolation was shifted towards higher silica concentrations at constant chain length or towards higher chain lengths at constant silica concentration by treating the silica surface. Hexamethyldisilazane was used to deactivate a partial number of silanol groups.
Finite clusters One of the most striking results concerns the mixture made from polymer chains of number-average molecular weight M n = 44.5 x 10a g mol- 1. The initial concentration of the mixture was Csi=0.17 (w/w). In the absence of any treatment of the silica surface, the residual amount of polymer Qr left bound to silica was 0.8 gg-1. Then, the values of Qr were 0.7, 0.4 and 0.4 g g- 1 for converted free silanol fractions equal to 0.17, 0.72 and 0.75, respectively. The corresponding partial crystallinity ratios were pc=0.32, 0 and 0, respectively (Table 3, Figure 4). Therefore, it is considered that loops formed from the adsorption process can collapse onto the silica surface when the density of free silanol groups is smaller than 0.3. Accordingly, they cannot participate in the crystallization process. Mixtures made from polymer chains characterized by a number-average molecular weight M n = 7.3 x 104 were also observed. Ratios of converted silanols were 0.17, 0.22 and 0.60; corresponding amounts of residual polymer were 1.0, 0.9 and 0.6 g g- 1, respectively. The partial crystallinity was found to be equal to 0.65, 0.68 and 0.57 (Table 3, Figure 4). Corresponding fractions of frozen bonds are reported in Figure 2.
POLYMER, 1990, Vol 31, April
747
Crystallization of s/l/ca-filled PDMS: D. Bordeaux and J. P. Cohen-Addad 1.0 0.8
~
nn
nl
N
b
0.6
m 0.4 ifiu
o
0.2 I
0
I
1.5 2.0 OR(W/W)
I
I
2.5
5,0
Figure 4 The partial crystallinity observed from mixtures of siloxane and silica with a chemical treatment of the surface. Experimental points are compared with the average curves obtained without any chemical treatment (Figure I). (a) The ratio of crystallinity ~si observed on PDMS chains. (b) The crystallinity Pc normalized with respect to that of corresponding pure polymer chains. Number-average chain molecular weights (10-4_~r,) are: 32.5 (4, within r-q), 7.3 (4, within <>) and 4.45 (4, within C)). Corresponding points without any treatment are also reported in Figure 1. Fractions of treatment are reported in Table 3
Infinite cluster Treated silica was also used to observe the formation of infinite clusters. The concentration of silica was kept constant, Csi=0.17(w/w), and the chain molecular weight was M, = 3.25 x 105. As long as an infinite cluster was formed, the partial crystallinity was characterized by pc_0.7. Then, for Q , < I , the partial crystallinity was found to decrease. Figures 1 and 4 clearly show that the crystallinity ratio is a constant in all silica-siloxane mixtures that form a gel although the silica concentration and the chain length may vary from one system to another. The fraction of trains (or of loops and tails) has a constant value whatever the amount of polymer bound per gram of silica. This property will also be discussed in a subsequent work. CONCLUSIONS The mechanical mixing of siloxane chains and silica particles induces an adsorption process for which char-
748
POLYMER, 1990, Vol 31, April
acteristic properties arise mainly from the full immersion effect of particles in the polymer melt: the amount of adsorbed polymer is proportional to the square root of the number-average molecular weight of chains. The purpose of the present work was to give a preliminary investigation of the structure of the adsorbed layers. The degree of freedom of monomeric units is determined by observing the partial crystallinity of the polymeric phase. Differential scanning calorimetry measurements performed after removing all free chains showed that about 10 monomeric units are frozen by the formation of one hydrogen bond between one silanol group and one oxygen atom of a chain skeleton. The site--bond percolation induced by polymer chains connecting silica particles was perceived from the state of partial crystallinity of the adsorbed layer. This preliminary study will be extended by an n.m.r. approach to the kinetics of crystallization of the polymeric layer 7. Measurements of differential scanning calorimetry performed on the adsorbed layer of silica-siloxane mixtures show clear evidence for the existence of a site-bond percolation transition of silica particles. The percolation process occurs through polymer molecules. The present work also shows that eight monomeric units on average are fixed in space by the formation of one hydrogen bond between a silanol group and a skeleton oxygen atom. The analysis of experimental results suggests that the area of silica surface available to polymer chains might vary the silica concentration. REFERENCES 1 Cohen-Addad, J. P., Roby, C. and Sauviat, M. Polymer 1985, 26, 1231 2 Cohen-Addad, J. P. Polymer 1989, 30, 1820 3 Schaefer, D. W. MRS Bull. 1988, XIII, 222 4 Viallat, A., Cohen-Addad, J. P. and Pouchleon, A, Polymer 1986, 27, 843 5 Stauffer, D. 'Introduction to Percolation Theory', Taylor and Francis, London, 1985 6 Cohen-Addad, J. P., Viallat, A. and Huchot, P. Macromolecules 1987, 20, 2146 7 Feio, G., Buntinx, G. and Cohen-Addad, J. P. J. Polym. Sci. 1989, 27, 1 8 Orrah, D. J., Semlyen, J. A., Dodgson, K. and Ross-Murphy, S. B. Polymer 1987, 28, 985 9 Clarson, J., Mark, J. E. and Dodgson, K. Polym. Commun. 1988, 29, 208